Method for preparing thin layers of ferroelectric material

Abstract

The present invention relates to a method for preparing a monodomain thin layer of ferroelectric material. The method includes implanting optical species into a first surface of a ferroelectric donor substrate to form an embrittlement surface and define a first layer between the embrittlement surface and the first surface of the substrate. The first surface of the donor substrate is assembled with a support substrate using a dielectric assembly layer, and the donor substrate is fractured at the embrittlement surface. The dielectric assembly layer (7b) has a lower hydrogen concentration than the first layer (3) or contains an oxide that prevents hydrogen diffusion into the first layer, or the dielectric assembly layer (7b) contains a barrier that prevents hydrogen diffusion. The preparation method also includes heat treatment of the free surface of the first layer to diffuse the hydrogen contained in the first layer and induce multidomain transformation of a surface portion of the first layer, followed by thinning the first layer to remove at least the surface portion.

Description

[Technical Field]

[0001] The present invention relates to a method for preparing thin layers of ferroelectric material, more particularly to a preparation method that allows maintaining the monodomain nature of the ferroelectric material in the thin layer of the final product, which is used, for example, in the fields of microelectronics, micromechanics, photonics, etc. [Background technology]

[0002] It should be noted in the preceding sentence that a ferroelectric material is a material that has an electric polarization in its natural state, which polarization can be reversed by applying an external electric field. A "ferroelectric domain" refers to each continuous region of the material in which the polarization is uniform (all dipole moments are aligned parallel to each other in a given direction). Thus, a ferroelectric material can be characterized as "monodomain" if the material is formed by a single region in which the polarization is uniform, or as "multidomain" if the ferroelectric material contains multiple regions with different polarizations.

[0003] Various methods are known from the state of the art for producing thin layers of ferroelectric material, for example using molecular beam epitaxy, plasma sputtering, laser pulse deposition or application of the SmartCut™ technology, in which a thin layer is removed from a solid substrate of ferroelectric material by breaking at a weak zone (or embrittlement plane) formed in the solid substrate by implantation of optical species.

[0004] The present invention more particularly relates to the preparation of thin ferroelectric layers obtained by applying such a method. Patent document 1 notes that ferroelectric layers of lithium tantalate transferred by a method involving hydrogen implantation lead to the formation of multi-domain layers. Such features affect the performance of devices formed on / in the thin layer, such as surface acoustic wave devices (SAW), making the layer unsuitable for use.

[0005] Patent Document 2 discloses the finishing of a thin ferroelectric layer transferred onto a support by an intermediate layer and the application of the SmartCut process. In one example, this finishing involves heat treatment of the thin layer followed by polishing. According to this document, this sequence can restore the initial monodomain properties of the thin ferroelectric layer. However, additional analyses conducted by the applicant showed that this finishing sequence alone cannot guarantee the monodomain nature of the layer throughout its entire thickness, i.e., from the interface with the intermediate layer to its free surface. More specifically, these analyses using PFM (piezoresponse force microscopy) measurements, generally performed over a depth of approximately 50 nm from the measurement surface, confirmed that the surface portion of the thin layer is indeed monodomain, while the buried portion juxtaposed with the intermediate layer is multidomain. To perform this "buried" PFM measurement, the layer was gradually thinned by chemical-mechanical polishing to approach the measurement surface of this intermediate layer.

[0006] Patent Document 3 also discloses a method for producing thin layers of ferroelectric material using the SmartCut technology. In this document, as in Patent Document 1, the layer to be removed and transferred onto a support substrate is prepared by exposing it to an electric field to improve or restore its monodomain ferroelectric properties. Such treatment requires, for example, the presence of a pair of electrodes placed on either side of the ferroelectric layer to be treated in electrical contact with the layer. Having such a pair of electrodes is not always possible, especially when the ferroelectric layer is transferred onto an electrically insulating substrate or layer. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] US20100088868 [Patent Document 1] FR3068508 [Patent Document 3] FR2914492 [Patent Document 4] FR2860341 [Patent Document 5] FR2933233 [Patent Document 6] FR2953640 [Patent Document 7] US2015115480 Summary of the Invention

[0008] The object of the present invention is to propose a method for preparing a thin layer of ferroelectric material that at least partially addresses the aforementioned drawbacks, and more particularly a method for transferring a layer of ferroelectric material that involves a step of implanting optical species, a method that allows the monodomain quality of the transferred layer to be preserved or restored throughout its thickness, so that the layer formation process does not require exposing the latter to an electric field to restore its monodomain quality, and allows it to be transferred onto an electrically insulating substrate or provided with a dielectric layer.

[0009] To this end, the subject of the present invention is a method for preparing a thin monodomain layer of ferroelectric material comprising the steps of: implanting optical species into a first surface of a ferroelectric donor substrate to form an embrittlement surface and defining a first layer between the embrittlement surface and the first surface of the donor substrate; assembling the first surface of the donor substrate to a support substrate using a dielectric assembly layer; fracturing the donor substrate at the embrittlement surface to transfer the first layer onto the support substrate and expose a free surface of the first layer; finishing the first layer, the finishing comprising heat treating the free surface of the first layer followed by thinning the first layer to form the thin monodomain layer; Includes.

[0010] According to the present invention, the dielectric assembly layer comprising an oxide has a lower hydrogen concentration than the first layer, or prevents diffusion of hydrogen into the first layer, or the dielectric assembly layer comprises a barrier that prevents diffusion of the hydrogen into the first layer.

[0011] Surprisingly, this precise sequence of preparatory steps, combined with the specific properties of the assembly layers, leads to the formation of a final thin layer with monodomain polarization throughout its thickness and with satisfactory crystal and surface quality.

[0012] According to other advantageous and non-limiting characteristics of the present invention, taken alone or in any technically feasible combination: The thinning includes a chemical mechanical polish applied to the free surface of the first layer.

[0013] The heat treatment is carried out at a temperature between 300° C. and the Curie temperature of the ferroelectric material constituting the first layer, for a period between 30 minutes and 10 hours.

[0014] The heat treatment is carried out in an oxidizing or neutral gaseous atmosphere.

[0015] The donor substrate comprises a ferroelectric material selected from LiTaO3 or LiNbO3.

[0016] The ferroelectric material has a crystal orientation between 30° RY and 60° RY.

[0017] The material of the support substrate is silicon.

[0018] The donor substrate is a block of solid material.

[0019] The donor substrate comprises a thick layer of ferroelectric material disposed on a manipulator substrate.

[0020] Here, the manipulator substrate has a thermal expansion coefficient that is the same as or close to that of the support.

[0021] The material of the first layer and the material of the support substrate have different coefficients of thermal expansion.

[0022] The support substrate includes a charge trapping layer.

[0023] The dielectric assembly layer comprises an oxide having nitrogen with a nitrogen / oxygen ratio of 0.01 or 0.05 or greater.

[0024] The dielectric assembly layer comprises silicon oxide with a nitrogen / oxygen ratio between 0.01 and 0.25, or between 0.05 and 0.01.

[0025] The dielectric assembly layer is formed by a stack including a first oxide layer disposed on a side of the support substrate and a silicon nitride layer disposed on a side of the first layer.

[0026] According to another aspect, the present invention provides a method for manufacturing a method of manufacturing a semiconductor device comprising: a thin, completely monodomain layer of ferroelectric material; a dielectric assembly layer in contact with the thin layer and including an oxide that prevents hydrogen diffusion into the thin layer or includes a barrier that prevents hydrogen diffusion into the first layer; a support in contact with the dielectric layer; The present invention proposes an apparatus including:

[0027] According to other advantageous, non-limiting features of this aspect of the invention, alone or in any technically feasible combination: the support substrate comprises a charge trapping layer disposed between the solid portion of the support and the dielectric assembly layer; The dielectric assembly layer comprises an oxide having nitrogen with a nitrogen / oxygen ratio of 0.01 or 0.05 or greater.

[0028] The dielectric assembly layer comprises silicon oxide with a nitrogen / oxygen ratio between 0.01 and 0.25, or between 0.05 and 0.01.

[0029] The dielectric assembly layer is formed by a stack including a first oxide layer disposed on a side of the support substrate and a silicon nitride layer disposed on a side of the first layer. [Brief explanation of the drawings]

[0030] Other features and advantages of the present invention will become apparent from the following detailed description of the invention, which description is given with reference to the accompanying drawings. [Figure 1] 1A is an explanatory diagram showing a first embodiment of the method according to the present invention; 1B is an explanatory diagram showing a first embodiment of the method according to the present invention; 1C is an explanatory diagram showing a first embodiment of the method according to the present invention; and 1D is an explanatory diagram showing a first embodiment of the method according to the present invention. [Figure 2] 2A is a diagram illustrating a second embodiment of the method according to the present invention; 2B is a diagram illustrating a second embodiment of the method according to the present invention; 2C is a diagram illustrating a second embodiment of the method according to the present invention; and 2D is a diagram illustrating a second embodiment of the method according to the present invention. [Figure 3] 3A and 3B are schematic illustrations of a method for preparing a layer according to the present invention; [Figure 4] 1 is an illustration of a surface acoustic device that can be manufactured using a method according to the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0031] To simplify the following description, the same reference symbols are used for identical elements or elements that perform the same function in different embodiments of the methods described.

[0032] The drawings are schematic and, for ease of reading, are not necessarily drawn to scale, in particular the thicknesses of the layers are not drawn to scale with respect to the lateral dimensions of these layers.

[0033] The expression "coefficient of thermal expansion" used in the remainder of this description with respect to a layer or substrate refers to the coefficient of expansion in a direction defined by the major plane defining this layer or substrate. If the material is anisotropic, the retained value of the coefficient is the value of maximum amplitude. The values ​​of the coefficient are measured at room temperature.

[0034] FIG. 4 shows a surface acoustic device that can be manufactured using the method according to the invention.

[0035] The thin ferroelectric layer 10 is arranged in direct contact with the support 7 via the dielectric layer 7b. Metal electrodes 11a, 11b are arranged on the thin ferroelectric layer 10. As is well known per se, a high-frequency electrical signal applied to one of the electrodes propagates in the thin ferroelectric layer 10 in the form of a surface wave, where it is processed (e.g. filtered) and recovered on the other electrode. The characteristics of the thin ferroelectric layer 10, and more generally of all the layers forming the stack in which the thin ferroelectric layer 10 is placed, are decisive for achieving the desired processing of the electrical signal, in particular the thickness and monodomain crystal quality of the thin ferroelectric layer 10.

[0036] The characteristics of the support 7 can affect the propagation of waves in the thin layer 10 due to acoustic coupling. Therefore, the nature and thickness of these layers are also crucial for achieving or at least influencing the desired processing of the electrical signal. In the example shown in FIG. 4, the support 7 includes a solid silicon portion 7a and a dielectric assembly layer 7b, preferably made of oxide, which is in direct contact with the thin ferroelectric layer 10. By way of example, the assembly layer 7b is made of silicon oxide or silicon nitride or is formed by a stack of layers composed of these materials. Preferably, the support 7 is electrically resistive to avoid electrical coupling with the high-frequency signals applied to the electrodes 11a, 11b, which could affect the propagation of the electrodes 11a, 11b. Therefore, the solid portion 7a can be formed by a silicon substrate with a high resistivity, i.e., a resistivity greater than 1000 ohm-cm, more preferably greater than 3000 ohm-cm.

[0037] To enhance the resistivity of the support 7, in the example of FIG. 4, provision is made for inserting a charge trapping layer 7c, for example made of polycrystalline silicon, between the solid portion 7a and the assembly layer 7b. The formation of such a layer and its role in contributing to the resistivity of the support substrate 7 are well known per se, and detailed descriptions thereof can be found in U.S. Pat. Nos. 5,629,999, 5,729,545, 5,769,629, and 5,769,545. Of course, this charge trapping layer 7c can be formed using techniques other than those providing a layer made of polycrystalline silicon. This layer may also contain carbon, or may consist of or include silicon carbide or an alloy of silicon and carbon. Alternatively, it is a matter of generating electric traps in the layer 7c by ion bombardment of the surface portion of the support with relatively heavy species (e.g., argon) to create crystalline defects therein that can trap charges. It is also possible to provide the charge trapping layer 7c made of a porous material by porosifying the surface portion of the substrate 7, for example, if the substrate 7 is made of silicon.

[0038] It should be noted that the device shown in Figure 4 does not provide an electrode below the thin ferroelectric layer 10. The presence of the dielectric assembly layer 7b does not allow contact with the back surface of the thin ferroelectric layer 10. Therefore, it is not possible to pole the thin ferroelectric layer 10 using the presence of such an electrode or the formation of an electrical contact, as described in the state of the art cited in the introduction to this application.

[0039] It is therefore important to have a method for preparing a thin ferroelectric layer 10 that can preserve or restore the monodomain character of this layer 10 without requiring the application of a repolarization voltage.

[0040] With reference to Figures 1 and 2, to be able to manufacture the device shown in Figure 4, the invention generally provides for transferring a first ferroelectric layer 3 onto a support substrate 7, the first layer 3 being taken from a monodomain ferroelectric donor substrate 1 by a transfer technique based on the implantation of light species such as hydrogen species.

[0041] According to this technique, hydrogen is implanted into the first surface 4 of the donor substrate 1 to form an embrittlement surface 2 embedded therein. A first layer 3 is thus defined between the embrittlement surface 2 and the first surface 4 of the donor substrate 1. The first surface 4 of the donor substrate 1 is then assembled with a support 7, in this case by a dielectric assembly layer 7b. The donor substrate 1 is then fractured at the embrittlement surface 2, for example, using a moderate heat treatment and / or the application of mechanical forces. The first layer 3 is then released from the donor substrate 1, exposing the free surface 8 of the first layer, while the other surface 4 of the thin layer is in direct contact with the assembly layer 7b of the support substrate 7.

[0042] To form a "useful" thin layer 10, it is generally necessary to provide steps for preparing the first transferred layer 3 thus transferred onto the support 7. These steps generally aim to improve the crystalline quality of the thin layer 10 and its surface condition (e.g., its roughness). These preparation steps may include a step of thinning the first layer 3 (e.g., by chemical-mechanical polishing, indicated as "CMP" in the table below) and / or a heat treatment step (for ferroelectric materials, this can be a heat treatment at around 500°C in a neutral atmosphere or an oxygen-containing heat treatment, indicated as "TTH" in the table below). Note that this step of heat treating the free surface 8 of the layer 3 is separate from the step that fractured the donor substrate. In fact, this heat treatment cannot treat the free surface, since the layer 3 has not yet been completely released during the fracture annealing.

[0043] However, applicant has observed that the method of preparing the first layer 3 transferred onto the support substrate 7 for the purpose of forming the thin layer 10 may result in the creation of multiple ferroelectric domains within the thin layer 10, giving it a multi-domain nature. As already mentioned, such a feature will affect the performance of devices formed on / in the transferred thin layer, such as a surface acoustic device as shown in FIG. 4, and therefore make the layer unsuitable for that use.

[0044] The present inventors have observed that the nature and sequence of the steps in preparing first layer 3 can significantly affect the ferroelectric properties of thin layer 10. From these initial observations, the present inventors have performed analyses of various sequences of preparation steps using piezoelectric force microscopy (PFM) techniques to characterize the monodomain or multidomain nature of the surface of thin layer 10. These experiments are summarized in the table below.

[0045] [Table 1]

[0046] It can be seen that the first layer 3 obtained at the end of the exfoliation step and before preparation exhibits a conformal ferroelectricity, i.e., monodomain (first row of the table). However, the surface state and crystalline quality of this layer require the application of a preparation step.

[0047] The fourth row of the table corresponds to the application of a method linking thinning and thermal treatment (CMP+TTH), and at the end of this treatment it can be seen that the thin layer has multi-domain ferroelectric properties.

[0048] Rows 2 and 3 represent the application of a single step of thinning CMP or thermal treatment TTH, respectively. Both of these steps lead to non-compliant properties of thin layer 3.

[0049] Applicant has recognized that by sequentially applying a thermal treatment step TTH followed by a thinning step CMP, the resulting thin layer 10 has a monodomain distribution of its ferroelectric polarization and acceptable crystal and surface quality, as observed on the surface of this layer by PFM.

[0050] Further analyses carried out by the Applicant have made it possible to identify that the formation of multiple ferroelectric domains in the thin layer 10 can be related to the presence of a strong hydrogen concentration gradient in the first layer 3 prior to its preparation. Indeed, it has been observed that the application of a heat treatment to a layer with such a gradient leads to the formation of multiple ferroelectric domains. The hydrogen implanted in the donor substrate 1 during the step of defining the first layer 3 is distributed in this substrate according to a profile with a concentration peak at the embrittlement surface 2. Thus, after fracturing, the first ferroelectric layer 3 transferred onto the support substrate 7 has, at its free surface 8, a hydrogen concentration profile of 10 21 at(atomic) / cm 3 This concentration decreases with the thickness of the thin layer in the direction of the substrate 7, reaching 10 19 at / cm 3 reach a level of the order of

[0051] If the preparation step consists of a first step of thinning the first layer 3 to the target thickness of the thin layer 10 followed by a heat treatment, the latter being applied while the thin layer still has a strong hydrogen gradient, particularly in the surface portion of the thin layer, this results in the transformation of the thickness or surface portion of the thin layer 10 where the hydrogen gradient is substantial into a plurality of ferroelectric domains.

[0052] By reversing these two steps, the surface portion of the first layer 3 with the strong hydrogen concentration gradient is transformed into a multi-domain ferroelectric layer. However, this multi-domain surface portion of the first layer 3 is removed in the subsequent thinning step, providing a thin layer 10 with the required monodomain properties.

[0053] It should be noted that the heat treatment, whether performed before or after the thinning step, diffuses the hydrogen with which the layer is saturated and reduces the thickness gradient of this layer. Therefore, heat treatments that may be applied later in the preparation of thin layer 10 are unlikely to affect the monodomain nature of this layer.

[0054] A very detailed analysis using piezoelectric force microscopy of thin layer 10 also revealed that after thinning, this thin layer 10 may have embedded multi-domain portions in contact with the assembly layer, which are not easily visible using simple PFM surface observation, where measurements are limited to a depth of approximately 50 nm.

[0055] From PFM observations of thin layers of several samples, applicants have determined that the properties of the dielectric assembly layer 7b play an important role with regard to the redistribution of hydrogen in the ferroelectric layer during the heat treatment of the method for preparing the thin layer 10.

[0056] If the assembly layer 7b is hydrogen-rich, as is the case with certain deposited oxides or dielectrics, it forms a hydrogen source that can cause migration during thermal treatment into the ferroelectric layer that the assembly layer 7b contacts. This hydrogen can accumulate in the ferroelectric layer, especially at the interface between this layer and the assembly layer 7b. This migration and accumulation can create a hydrogen concentration gradient that allows for multi-domain transformation of the portion of the ferroelectric layer that is in contact with the assembly layer during thermal treatment.

[0057] Conversely, if assembly layer 7b is relatively hydrogen-depleted, i.e., has a lower hydrogen concentration than that present in the transferred ferroelectric layer, excess hydrogen in the ferroelectric layer will be absorbed into assembly layer 7b during diffusion by the thermal treatment, thus preventing hydrogen accumulation at the assembly interface and avoiding multi-domain transformation in the portion of the ferroelectric layer near this interface.

[0058] The present invention takes advantage of these results and observations to propose a method for preparing a thin layer 10. More specifically, the present invention relates to a method for preparing a thin layer 10 of ferroelectric material that is transferred from a donor substrate 1 to a support substrate 7 by a transfer technique involving the implantation of photospecies such as hydrogen species, as described very generally above. Several embodiments exist for carrying out this transfer.

[0059] According to a first embodiment shown in FIGS. 1A to 1D, the donor substrate 1 is a solid monodomain block of ferroelectric material, such as LiTaO3, LiNbO3, LiAlO3, BaTiO3, PbZrTiO3, KNbO3, BaZrO3, CaTiO3, PbTiO3, or KTaO3. The donor substrate 1 can take the form of a circular wafer of standardized size, such as 150 mm or 200 mm in diameter. However, the present invention is in no way limited to these dimensions or this shape. The donor substrate 1 may have been removed from an ingot of ferroelectric material, and this removal was performed so that the donor substrate 1 has a predetermined crystal orientation. The orientation is selected depending on the intended application. Thus, when the objective is to utilize the properties of thin layers to form a SAW filter, it is common practice to select an orientation between 30° R-1 and 60° R-1 or between 40° R-1 and 50° R-1. However, the present invention is in no way limited to a specific crystal orientation.

[0060] Regardless of the crystalline orientation of the donor substrate 1, the method involves the introduction of hydrogen into this donor substrate 1. This introduction can correspond to an implantation of hydrogen, i.e. an ion bombardment of the plane 4 of the donor substrate 1 with hydrogen. Naturally, this introduction of hydrogen, for example by implantation, can be expected to be complemented by the introduction of other photospecies, such as helium.

[0061] In a manner known per se, as shown in FIG. 1B, the purpose of the implanted ions is to form an embrittlement plane 2 that demarcates a first layer 3 of the ferroelectric material to be transferred flanking the face 4 and another portion 5 that forms the remainder of the substrate.

[0062] The nature, dose and implantation energy of the implanted species are chosen as a function of the thickness of the layer intended to be transferred and of the physicochemical properties of the donor substrate 1. Thus, in the case of a donor substrate 1 made from LiTaO3, an energy between 30 keV and 300 keV is chosen to delimit the first layer 3 on the order of 200 nm to 2000 nm. E 16 and 5E 17at / cm 2 It is possible to select a hydrogen dose between 0.1 and 0.2.

[0063] In the next step, shown in FIG. 1C, the flat surface 4 of the donor substrate 1 is assembled with the surface 6 of the support substrate 7. The support substrate 7 may be the same size and shape as the donor substrate 1. For reasons of availability and cost, the support substrate 7 is a silicon, monocrystalline, or polycrystalline wafer. However, more generally, the support substrate 7 can be formed of any material, for example, silicon, or even an electrically insulating material such as sapphire or glass, and may be of any shape. To form the device shown in FIG. 4, a support substrate is selected that is formed by a solid portion 7a made of highly resistive silicon with a charge trapping layer 7c made of polycrystalline silicon, for example.

[0064] Prior to the assembly process, it is possible to consider preparing the surface of the substrate to be assembled using cleaning, brushing, drying, polishing processes, or using, for example, oxygen or nitrogen based plasma activation.

[0065] The assembly process can involve intimate contact between the donor substrate 1 and the support substrate 7 by molecular adhesion and / or electrostatic bonding. In order to facilitate the assembly of the two substrates 1, 7, provision is made for forming at least one dielectric assembly layer 7b on the face 6 of the support substrate 7 to be assembled. It is also or alternatively possible to form an assembly layer on the face 4 of the donor substrate 1 to be assembled, preferably before the formation of the embrittlement face 2. In other words, the dielectric assembly layer 7b can be formed, at least in part, on one and / or the other of the two substrates 1, 7.

[0066] The dielectric assembly layer 7b is formed, for example, by silicon oxide, or more generally by an oxide such as Ta2O5, ZrO2 or HfO (these oxides, like SiO2, can be formed by deposition) or silicon nitride, and has a thickness between a few nanometers and a few microns. The assembly layer 7b can consist of a stack of different types of dielectric layers. The dielectric assembly layer 7b can be produced by various well-known techniques, for example, by thermal oxidation or nitridation processes, chemical deposition (PECVD, LPCVD, etc.).

[0067] Generally, deposition or formation techniques for the assembly layer are preferred, leading to the incorporation of small amounts of hydrogen therein. It should be noted that the choice of material forming this assembly layer, its thickness, and its formation technique are not entirely free. This choice is determined, inter alia, by the electrical insulation, acoustic wave propagation, and / or adhesion properties of this layer in the final structure, as well as by the intended application. Finally, it should be noted that forming the assembly layer by deposition, especially oxide deposition, is generally easier and more versatile, since this formation method can be carried out independently of the nature of the support and at relatively moderate temperatures, sometimes lower than 1000°C or 950°C, or even lower than 900°C, which may be imposed by the nature of the solid part of the support 7a or by the presence of the trapping layer 7c.

[0068] According to an important feature of one embodiment, care is taken to ensure that the dielectric assembly layer 7b has a lower hydrogen concentration than the average hydrogen concentration of the first layer 3. By assuming that the majority of the hydrogen in the first layer 3 comes from the hydrogen implanted to form the embrittlement surface 2, the average concentration of hydrogen is determined by the implanted dose (at / cm 2 ) by the thickness (cm) of the first layer 3. 20 at / cm 3 and 10 22 at / cm 3By way of example, if the assembly layer 7b is formed by silicon oxide deposited on a support substrate, this oxide is annealed before the assembly step so as to exodiffuse most of the hydrogen it contains.

[0069] Generally, the step of annealing assembly layer 7b can be provided to subject this dielectric layer to a higher temperature than the heat treatment for preparing the first transfer layer described later in this description. This layer can therefore be subjected to temperatures of 600°C, 700°C, or even 800°C or higher. The average hydrogen concentration in assembly layer 7b after this exodiffusion step is therefore 5.10 20 at / cm 3 less than, or advantageously less than 10 18 at / cm 3 less than, or more preferably 10 18 at / cm 3 It should be noted that annealing can modify characteristics of the assembly layer other than the hydrogen concentration. In particular, it can reduce the diffusivity of hydrogen, i.e., the ability of this species to diffuse within the materials forming the assembly layer, so that hydrogen can be present at relatively high concentrations (10 20 at / cm 3 Even if the amount of oxygen is small (approximately 100%), the possibility of the oxygen diffusing toward the first layer 3 is low.

[0070] Alternatively, or in addition to this embodiment, provision can be made for providing the dielectric assembly layer 7b with a barrier layer that prevents the diffusion of hydrogen into the first layer 3. By way of example, the assembly layer 7b can be formed by a stack formed by a first layer of oxide, for example silicon oxide, arranged on the side of the support substrate 7a (or, in the latter case, the trapping layer 7c), and a layer of silicon nitride arranged on the side of the first ferroelectric layer 3. This nitride layer prevents hydrogen that may be present in the silicon oxide layer from diffusing towards the first ferroelectric layer 3.

[0071] According to yet another approach, the dielectric assembly layer can be provided with any concentration of hydrogen, but the layer has very low diffusivity of this hydrogen, and therefore remains sufficiently confined to not significantly diffuse in the direction of the thin layer 3.

[0072] Such layers can be formed by oxides, such as silicon oxide (SiON), with nitrogen-to-oxygen ratios of 0.01, 0.05, or even 0.1 or greater. When the assembly dielectric layer 7b is based on silicon oxide (SiON), this material is very common because it has well-known properties for acoustic wave formation, adhesion, electrical insulation, and propagation. To avoid excessively altering these functions and maintain performance comparable to or close to that of simple silicon oxide (SiO2), it is possible to determine not to exceed a nitrogen-to-oxygen ratio of 0.1 or 0.25. It should be noted that such nitrogen-rich oxide layers can be easily formed by deposition techniques, such as PECVD, in which at least one of the carrier gases can be selected to be nitrogen, which can be incorporated into the oxide layer in a controlled manner. The nitrogen-to-oxide ratio can be established by measurements using a technique called EDX ("energy dispersive X-ray spectroscopy") or by measurements of nitrogen and oxygen measured in SIMS (secondary ion mass spectroscopy) measurements of the oxide layer 4.

[0073] The dielectric assembly layer 7b may be formed throughout its thickness from a material with low diffusivity, for example a nitrogen-containing oxide as described above.

[0074] Alternatively, as explained in the previous alternative, only a barrier layer can be provided to prevent the diffusion of hydrogen into the first layer 3 made of this material with low diffusivity. For example, a layer of silicon oxide with a high concentration of hydrogen can be formed by deposition on a support substrate, this layer being prepared using a nitrogen-based plasma to incorporate this nitrogen superficially. A barrier layer of nitrogen-rich SiO is thus formed at this superficial thickness, preventing the diffusion of hydrogen contained in the remaining oxide.

[0075] If the solid part of the support 7a is provided with a charge trapping layer 7c comprising silicon at its surface, it is conceivable to form the dielectric assembly layer 7b by thermally oxidizing the surface part of this trapping layer by treating it in an oxidizing atmosphere at a moderate temperature below 950°C, preferably between 800°C and 900°C. This therefore avoids damage to the charge trapping layer 7c due to, for example, recrystallization, caused by exposing it to high temperatures above 950°C or 900°C. Optionally, a step of polishing the surface of the trapping layer 7c thus oxidized can be introduced to make it compatible with subsequent assembly steps.

[0076] Regardless of the nature of the dielectric assembly layer 7b and the treatment it has undergone, at the end of this assembly step, an assembly is provided as shown in FIG. 1C, comprising two associated substrates: the flat surface 6 of the support substrate 7 adhered to the flat surface 4 of the donor substrate 1 by the dielectric assembly layer 7b.

[0077] The assembly is then treated to release the first layer 3 of ferroelectric material from the donor substrate 1 by cleaving at the embrittlement plane 2 .

[0078] This peeling step may therefore comprise applying a heat treatment to the assembly within a temperature range of the order of 80° C. to 300° C., in order to enable the first layer 3 to be transferred onto the support substrate 7. Instead of or in addition to the heat treatment, this step may comprise the application of a blade, or a jet of gas or liquid fluid, or any other force of a mechanical nature, at the embrittlement surface 2.

[0079] Following this separation step, the structure 9 shown in Figure 1D is obtained, which comprises a first layer 3 of ferroelectric material with a first free face 8 and a second face 4 (corresponding to the first face of the donor substrate) disposed on a support substrate 7, with a dielectric assembly layer 7b disposed between the first layer 3 and the remainder of the support 7.

[0080] 2A to 2D show a second embodiment that is particularly suitable for producing heterogeneous structures 9, in which the first layer 3 has a thermal expansion coefficient that is very different from that of the support, for example with a difference of more than 10%.

[0081] Referring to FIG. 2A, the donor substrate 1 in this case consists of a thick layer of ferroelectric material 1a and a manipulator substrate 1b, with the same properties as described for the solid block of ferroelectric material in relation to the first embodiment.

[0082] The manipulator substrate 1b is advantageously formed from a material (or materials) that provides a coefficient of thermal expansion close to that of the support substrate 7. "Close" means that the difference in the coefficient of thermal expansion between the manipulator substrate 1b and the support is smaller in absolute value than the difference in thermal expansion between the solid block of ferroelectric material and the support substrate 7.

[0083] Preferably, the manipulator substrate 1b and the support substrate 7 have the same thermal expansion coefficient. During assembly of the donor substrate 1 and the support substrate 7, a structure is formed that can withstand heat treatment at relatively high temperatures. For ease of implementation, this can be achieved by selecting the manipulator substrate 1b to be made of the same material as the support substrate 7.

[0084] To form the donor substrate 1 of this embodiment, a solid block of ferroelectric material is first assembled with the manipulator substrate 1b, for example, according to the molecular adhesion bonding technique described above. Next, a layer of ferroelectric material 1a is formed by thinning, for example, by grinding and / or chemical-mechanical polishing and / or etching. Prior to assembly, one and / or the other of the contacted surfaces can be prepared to form an adhesion layer (for example, by depositing silicon oxide and / or silicon nitride). The assembly can include the application of a low-temperature heat treatment (for example, 50°C to 300°C, typically 100°C) to allow sufficient strengthening of the bonding energy to enable the subsequent thinning step.

[0085] The manipulator substrate 1b is selected to have a thickness substantially equal to that of the support substrate 7. The thinning step is performed so that the thickness of the thick layer 1a is thin enough that the stresses generated during the heat treatments applied in the remaining part of the method have a lower intensity. At the same time, this thickness is sufficient to allow the first layer 3, or several such layers, to be removed. This thickness may be, for example, between 5 and 400 microns.

[0086] The following steps of the process of this second implementation are equivalent to those described in the first embodiment. As shown in FIG. 2B, hydrogen is implanted into the thick layer 1a to create an embrittlement surface 2, which indicates the separation of the first layer 3 from the remaining part 5 of the donor substrate 1. This step is followed by the assembly of the donor substrate 1 on the support substrate 7, as shown in FIG. 2C. A dielectric assembly layer 7b is provided between the donor substrate 1 and the support substrate 7, and this assembly layer has the same properties as provided in the context of the first embodiment. If it is decided to form a layer of silicon oxide (or a dielectric of another nature that requires exodiffusion annealing of hydrogen) on the face of the assembled donor substrate 1 of this embodiment, its thickness is limited to a few nanometers, for example, less than 10 or 50 nm. This is because it may be difficult to apply exodiffusion annealing to this layer without risking damage to the donor substrate 1, which is composed of an assembly of materials with significantly different thermal expansion coefficients.

[0087] The first layer 3 is then separated from the rest of the substrate 5 to obtain the structure 9 shown in FIG. 2D.

[0088] This embodiment is advantageous in that the assembly formed from the donor substrate 1 and the support 7 can be subjected to temperatures much higher than those applied in the first embodiment without the risk of uncontrolled destruction of one of the substrates or delamination of the donor substrate 1. That is, the balanced structure in terms of the thermal expansion coefficients of this assembly makes it possible to facilitate the step of separating the thin layer 3 by subjecting the assembly to relatively high temperatures, for example between 100°C and 600°C.

[0089] Regardless of the embodiment selected, and as specified above, a step of preparing the first layer 3 is then required to form a thin layer 10 with satisfactory crystallographic and surface quality.

[0090] As shown in FIG. 3A, this method first involves heat treatment of the free surface 8 of the transferred first layer 3. This heat treatment can correct crystal defects present in the first layer 3 and also helps to strengthen the bond between this layer 3 and the support 7. The above research also shows that if the heat treatment is performed at a sufficient temperature, it can have the effect of causing diffusion of hydrogen contained in the first layer 3, particularly in its surface portion, and causing multi-domain formation in this surface portion.

[0091] The surface portion is about 50 nm thick or less and can be established throughout the entire extent of layer 3. At the end of the heat treatment, ferroelectric layer 10 has a relatively constant concentration of hydrogen throughout its thickness, 10 18 at / cm 3 and 10 19 at / cm 3 It should be noted that due to the properties of assembly layer 7b described above, the hydrogen contained in this assembly layer is less likely to diffuse towards thin layer 3 during annealing and / or to be trapped at the interface that exists between assembly layer 7b and this layer 3. This therefore avoids the formation of a hydrogen concentration gradient in this buried zone near this interface, thus preventing this zone from being transformed into a multi-domain property.

[0092] The heat treatment for preparing the compliant thin layer 10 involves subjecting the structure to a temperature between 300°C and the Curie temperature of the ferroelectric material (preferably 450°C, 500°C, or 550°C or higher to promote hydrogen diffusion) for between 30 minutes and 10 hours. The heat treatment can provide a gradual gradient. This heat treatment is preferably carried out by exposing the free surface 8 of the first layer 3 to an oxidizing or neutral gaseous atmosphere, i.e., without covering this surface of the thin layer with a protective layer capable of preventing hydrogen exodiffusion.

[0093] 3B, the preparation method also includes thinning the thin layer 3 after the heat treatment. This thinning can correspond to polishing the free surface 8 of the thin layer 3, for example, by mechanical, chemical, mechanical thinning techniques and / or chemical etching. It allows the multi-domain surface portion of the thin layer 10 to be removed, with the free surface 8 being prepared to exhibit a low roughness, for example, by atomic force measurement (AFM), of less than 0.5 nm RMS 5 × 5 μm. To reach the target thickness of the thin layer 10, and in all cases a thickness greater than the thickness of the multi-domain surface portion, a removal of a thickness of 100 nm to 300 nm is generally provided.

[0094] This therefore constitutes a thin monodomain layer throughout its thickness, with the necessary surface state quality for crystal quality. The dielectric assembly layer 7b, which has a lower hydrogen concentration than the first layer 3 or contains a barrier to hydrogen diffusion into the first layer 3, avoids the accumulation of excess hydrogen in the buried zone of the thin layer 10 located near the assembly interface. This prevents a multidomain transformation in this zone, which would be impossible to eliminate.

[0095] Of course, the invention is not limited to the described embodiments and examples, and modified embodiments can be provided without departing from the scope of the invention as defined by the claims.

Claims

1. A method for preparing a thin monodomain layer (10) of ferroelectric material, comprising: implanting hydrogen ions into a first surface (4) of the donor substrate (1) made of a ferroelectric material to form an embrittlement surface (2), and defining a first layer (3) between the embrittlement surface (2) and the first surface (4); assembling the first side (4) of the donor substrate (1) to a support substrate (7) using a dielectric assembly layer (7b), the dielectric assembly layer (7b) being annealed before the assembling step; - fracturing the donor substrate (1) at the embrittlement surface (2) to transfer the first layer (3) to the support substrate (7) and to expose a free surface (8) of the first layer (3); finishing the first layer (3), which comprises heat treating the free surface (8) of the first layer (3) to convert a surface portion of the first layer (3) to have a plurality of ferroelectric domains, and subsequently thinning the first layer (3) to remove the surface portion having the plurality of ferroelectric domains and form a thin monodomain layer (10); Including, the dielectric assembly layer (7b) comprises an oxide that prevents the diffusion of the hydrogen into the first layer, or the dielectric assembly layer (7b) comprises a barrier that prevents the diffusion of the hydrogen towards the first layer (3), A method characterized in that the annealing before the assembling step is carried out at a higher temperature than the heat treatment in the finishing step of the first layer (3).

2. 2. A method according to claim 1, characterized in that said thinning comprises a chemical mechanical polishing applied to said free surface (8) of said first layer (3).

3. 3. A method according to claim 1 or 2, characterized in that the heat treatment is carried out at a temperature between 300°C and the Curie temperature of the ferroelectric material constituting the first layer (3), for a period between 30 minutes and 10 hours.

4. 4. The method according to claim 1, wherein the heat treatment is carried out in an oxidizing or neutral gaseous atmosphere.

5. The donor substrate (1) is LiTaO 3 or LiNbO 3 5. The method of claim 1, further comprising the step of:

6. 6. The method of claim 1, wherein the ferroelectric material has a crystal orientation between 30° and 60° RY.

7. 7. A method according to any one of claims 1 to 6, characterized in that the material of the support substrate (7) is silicon.

8. 8. A method according to any one of claims 1 to 7, characterized in that the donor substrate (1) is a block of solid material.

9. 8. A method according to any one of claims 1 to 7, characterized in that the donor substrate (1) comprises a thick layer (1a) of the ferroelectric material arranged on a manipulator substrate (1b).

10. 10. The method according to claim 9, wherein the manipulator substrate (1b) has a thermal expansion coefficient that is the same as or close to the thermal expansion coefficient of the support substrate (7).

11. 11. A method according to any one of the preceding claims, characterized in that the material of the first layer (3) and the material of the support substrate (7) have different thermal expansion coefficients.

12. 12. A method according to any one of the preceding claims, characterized in that the dielectric assembly layer (7b) comprises an oxide with nitrogen in a nitrogen / oxygen ratio of 0.01 or more.

13. 13. A method according to any one of the preceding claims, characterized in that the dielectric assembly layer (7b) comprises silicon oxide with nitrogen in a nitrogen / oxygen ratio between 0.01 and 0.

25.

14. 14. The method according to claim 1, wherein the dielectric assembly layer (7b) is formed by a stack including a first oxide layer arranged on the side of the support substrate (7) and a layer of silicon nitride arranged on the side of the first layer (3).

15. a thin monodomain layer (10) of ferroelectric material; a support substrate (7) in contact with said thin monodomain layer (10); Equipped with the support substrate (7) comprises a dielectric assembly layer (7b), the dielectric assembly layer (7b) being in contact with the thin monodomain layer (10) and comprising an oxide that prevents hydrogen diffusion into the thin monodomain layer (10); A device characterized in that said dielectric assembly layer (7b) has a hydrogen concentration that is lower than the average hydrogen concentration of said thin monodomain layer (10).

16. 16. The device according to claim 15, characterized in that the support substrate (7) comprises a charge trapping layer (7c) arranged between the solid part (7a) of the support substrate (7) and the dielectric assembly layer (7b).

17. 17. Device according to claim 15 or 16, characterized in that the dielectric assembly layer (7b) comprises an oxide with nitrogen in a nitrogen / oxygen ratio of 0.01 or more.

18. 18. A device according to any one of claims 15 to 17, characterized in that the dielectric assembly layer (7b) comprises silicon oxide with nitrogen in a nitrogen / oxygen ratio of between 0.01 and 0.

25.

19. 19. The device according to any one of claims 15 to 18, characterized in that the dielectric assembly layer (7b) is formed by a stack including a first oxide layer arranged on the side of the support substrate (7) and a layer of silicon nitride arranged on the side of the thin monodomain layer (10).

Images